I've sat through a lot of adtech interviews. A lot. Most of them are the verbal equivalent of a PowerPoint deck that could have been an email. A bunch of guys in Patagonia vests telling you the future is "exciting" while saying absolutely nothing that would help a single human being make a single better decision. It's like watching two people aggressively agree with each other for 45 minutes and calling it "thought leadership."

So when Mike Shields, one of the genuinely good interviewers in this space, maybe the best, a guy who actually interrupts people and says "wait, explain that like I'm a normal person," sat down with Charles Manning at the Kochava Summit in Sandpoint, Idaho, I paid attention.

And you should too. Because what Manning laid out isn't the usual "AI is going to change everything" fortune cookie nonsense. It's a specific, uncomfortable, and frankly jarring timeline for how fast advertising is about to get completely rearranged. Not in a decade. Not in five years.

In about sixteen months.

Let that marinate.

Shields opened with something I think a lot of us are feeling but are too polite or too exhausted to say out loud: he's got metaverse PTSD. And honestly? Same.

Remember that whole fever dream? Remember when every conference had a "metaverse track" and we were all supposed to be buying virtual sneakers in Decentraland by 2024? Remember when Mark Zuckerberg renamed his entire company after a concept that turned out to be a glorified Wii Sports lobby with worse graphics and no friends? Remember when people were spending real actual money on virtual real estate? In a recession? I remember. My therapist remembers.

So yeah, when someone walks up to a microphone and says "agentic advertising is going to transform everything," the very reasonable instinct is to smile politely, say "sure, buddy," and go back to your eleven-tab spreadsheet that nobody asked for but everyone depends on.

Manning's response was essentially: I hear you. But you're dead wrong to compare them.

The metaverse was a consumer bet. It required billions of human beings to voluntarily change their behavior, strap screens to their faces, and pretend to enjoy virtual meetings where their avatar's legs didn't render. Agentic AI is not asking consumers to do a single thing. It's changing the plumbing. It's changing how the work gets done behind the curtain, whether the people doing that work have updated their LinkedIn headline to "AI-curious" or not.

And that is a fundamentally different kind of disruption. One requires mass adoption. The other just requires your boss to realize the team of twelve can be a team of four with better output.

Fun times.

Manning dropped this line almost casually and I had to rewind it because it's the single cleanest framing I've heard of what's actually happening. I'm going to put it in bold because it deserves it:

"Programmatic is to the auction as agentic is to the workflow."

That's a sentence that deserves to be tattooed on the forearm of every agency holding company CEO who's currently pretending they have an AI strategy. Programmatic changed how ads get bought and sold. Agentic changes how the people buying and selling ads actually spend their day. The first one disrupted the transaction. This one disrupts the job description.

Shields, to his credit, pushed on this. Okay, so is this "just" making workflow better? Making reports faster? Letting people spend more time on "strategy" (the word every consultant uses when they mean "the stuff we can't automate yet")? Or is this something bigger?

Manning basically said: yes.

Not "yes to one of those." Yes to all of it. At the same time. At a speed that is going to make people's heads spin.

Shields asked the obvious follow-up: is this going to be like the programmatic shift? That took, what, a decade to really play out? There was a gradual buildup, then an explosion, then a long messy middle where everyone argued about supply path optimization and pretended they understood header bidding.

Manning basically laughed. "Breathtakingly faster" is what he said.

And here's what makes this genuinely, structurally different from every other hype cycle I've lived through and written about. The Lego blocks already exist. The APIs are built. The infrastructure is there. The pipes are laid. What's happening now isn't building new railroad tracks across the frontier. It's someone realizing there's a bullet train sitting on tracks that were laid ten years ago and somebody just figured out how to turn the thing on.

MCP (model context protocol, for the normal humans in the audience) is the layer that lets AI models actually do things through existing APIs. Manning described it as basically the API layer for AI and LLMs. Your AI agent doesn't need a special new integration. It just needs to talk to the APIs that already exist, that your company already built, that your team already sort of uses but mostly ignores because pulling data from them is annoying.

Now it's not annoying anymore. Now it's automatic. And that's the ballgame.

Here's where Manning got really specific and, honestly, a little scary. He walked through a concrete example. Today, right now, in most organizations, after a campaign runs, somebody (usually a junior person who deserves a raise) spends about a week pulling together data from multiple spreadsheets, multiple platforms, multiple sources to create a post-campaign brief. This brief then gets delivered roughly 45 days after the campaign ended. And then decisions get made based on it.

Read that again. Forty-five days. In an industry that prides itself on "real-time." We've been buying impressions in milliseconds and evaluating whether they worked a month and a half later. That's like running a Formula 1 race and getting your lap times by postal mail.

Manning's argument, and I think he's right, is that with LLMs and even smaller language models (SLMs), that entire six-step workflow can now happen daily. Not weekly. Not monthly after everyone's forgotten what the campaign was even for. Daily.

The signal has been real-time for years. Now the process catches up to the signal. And if that doesn't make the hair on the back of your neck stand up a little, you're either not paying attention or you're the person who's about to be replaced by a workflow that runs while you sleep.

I'm going to get deeper into the specific implications for agencies, for brands, for the adtech daisy chain, and for CTV in the coming parts of this series. But I want to leave you with the thing that stuck with me most from this interview.

Manning said the most valuable skill in engaging with AI is the ability to articulate your desire. Not data science. Not programming. Not the ability to build a pivot table that would make your manager weep. The ability to read, write, and communicate clearly.

Shields, bless him, did a little fist pump at this because, like me, he's a words guy who has spent the last decade being told he needs to learn Python or die. And look, I'm not saying data science is dead. But Manning's point is profound: when the machine can do the technical execution, the human advantage becomes clarity of thought and communication.

The liberal arts kids might actually win this one. Which, if you spent four years getting a comparative literature degree and then being mocked at every Thanksgiving dinner by your cousin who went into finance, is deeply, cosmically satisfying.